Table 5.

Limitations of the current literature and opportunities for future work

Limitation	Opportunity
1. Relatively small retrospective cohorts with limited events	Establishing multi-centre networks for future larger-scale studies
2. Multiple segmentation techniques used	Consensus on segmentation technique for MTV and TLG and development of automated AI-methods which are implemented within reporting software by manufacturers
3. Single site models using a single dataset	Internal and external validation should be routinely performed and facilitated by networks
4. Varying predictive end points	Consensus on clinically relevant predictors
5. Small numbers of papers using non-linear analysis	Using different machine learning and deep learning models to aid in imaging analysis and outcome prediction